Analyzing the Effect of Large Language Models on the Classification of Vehicle Social Interactions

Institut

Lehrstuhl für Ergonomie

Typ

Bachelorarbeit / Semesterarbeit / Masterarbeit /

Inhalt

experimentell / konstruktiv /  

Beschreibung

Driver interaction plays a critical role in road traffic, especially in dense flow conditions. With the increasing presence of automated vehicles (AVs), it is important to understand how current human-driven interactions may change and how AVs can be integrated safely and effectively. This thesis focuses on classifying naturalistic interactions between drivers, as observed in a connected driving simulator study using two simulators. The goal is to identify common types of interactions and use various large language models (LLMs) to categorize these interactions. The findings will help assess how AVs can better adapt to mixed traffic conditions involving both human drivers and autonomous systems.

 

The work is divided into four main phases:

• Literature Review: Analyze existing studies on vehicle interaction, prompt engineering for LLMs, and subjective evaluation methods.
• Expert Workshop: Conduct structured workshops to derive ground truth labels for driver interactions.
• Classification: Apply various LLMs and prompt strategies to classify vehicle interactions.
• Evaluation: Compare and validate the results of different LLM-based approaches against expert judgments.

 

Voraussetzungen

• Background in Human Factors, Mechanical Engineering, Computer Science, Artificial Intelligence, or related fields.
• Interest in driver behavior, autonomous driving technologies, and the application of large language models.
• Solid knowledge of statistics is an advantage.
• Independent, structured, and analytical working style.
• Good command of German or English, and strong communication skills for conducting expert interviews.

 

If you are interested, please send a brief application including a short motivation statement (3–4 sentences), your CV, and a current transcript of records to tianyu.tangtum.de. I look forward to receiving your application.

 

Tags

Lfe Tang

Möglicher Beginn

sofort

Kontakt

Tianyu Tang, M.Sc
Raum: MW 3328
Tel.: 01731987261
tianyu.tangtum.de